chembal.com

You’ve probably seen similar clocks to the one I made in this project in gift catalogs and online. They were super expensive at first, though I’ve seen knock-offs on eBay recently for as cheap as $25, and some are probably even less. The idea behind it is to have what appears to be just a solid block of wood display the time mysteriously from it. I suspect the cheaper versions of this probably just wrap a cheap LED clock with a wood veneer to achieve the effect cheaply, and it probably works just as good as my expensive version. But, my goal wasn’t to just re-create the effect, I wanted to machine my clock out of one solid block of maple. On my fifth try, I am finally satisfied with the results, and I present it here for your enjoyment.

Block Design

It seems simple enough.  Take a chunk of maple, mill out some pockets for digit segments, the electronics, and a back panel.  Drill and tap a few holes so everything goes together.  The tricky part of the project, however, is to consistently pocket digit segments to within about 20 thousandths of the front face of the block.  This works just fine on a perfectly set up machine with a perfectly flat block of wood.  The challenge, I found, was my machine wasn’t perfectly set up, and you can either have a perfectly flat block of wood or a nicely hand sanded and finished block of wood, but not both.

• Attempts 1 and 2: My first blocks of maple ended up with nice thicknesses at one end, but the digits were too thick at the other.  After a few sleepless nights and lots of fiddling, I found that my Z-axis encoder bushing was out of alignment, causing Z-axis drift.  Re-assembling the encoder ceased the drift.
• Attempt 3:  This time I was over-zealous with the pocketing depth.  I had noticed that the digits that worked on previous attempts weren’t bright enough to my satisfaction, and I over-compensated.  The result was a neatly perforated, and completely useless, block of maple.
• Attempt 4:  This time I machined it right.  I had re-faced my spoil-board and both sides of the maple, and I was careful about the pocketing depth.  After assembling and installing my clock PCB, I was thrilled to see the time neatly shining through.  The joy was brief.  When I sanded and applied a Polyurethane finish to the maple, the thin segments absorbed different amounts of finish than the thicker parts of wood, causing the ultra-thin segments to wrinkle.  The clock still worked and looked kind of neat from a distance, but the effect was ruined for me.  When the clock was off, it was still obviously a seven-segment display instead of a nice block of wood.  I had already run through my whole expensive chunk of 10/4 maple and many hours of work on this thing but I wasn’t going to call this good enough.  I needed a new plan.
• Attempt 5:  At this point, it was clear that I needed to start with a finished block of wood, and machine it last, to avoid an inconsistent look on the front of the block.  I bought another huge chunk of 10/4 maple and prepared to get this done right.  I started by cutting a chunk big enough for two clocks, rough sized the two pieces on the table saw, and was left with an inch and a half wide chunk for testing.  I faced them all parallel and to size, sanded them, and finished five sides on each with three coats of spray lacquer with light sanding in between coats.  At this point, they were magnificent, smooth blocks of maple.  I modified my G-Code to test my narrower, but otherwise identical to the others, test block.  The test results revealed that despite my best efforts, the block was 15 thousandths thinner in the middle than the sides.  This was hugely disappointing.  If I calibrated for the sides, the center would punch through.  If I calibrated for the center, the sides would be too dim.  No uniform depth would produce a useful display.  So, I decided to adjust for it.  I changed my tool path to go .025″ short of the spoil-board on the sides, increasing by .005″ at a time up to .040″ short of the spoil-board at the center.  I wasn’t sure how this would look, but I ran with it.  After a few hours staring at my machine turning most of my block of maple into sawdust, I was successful.  I installed my clock PCB, and was overjoyed to see just the image I had imagined seeing during the whole run of the project.  The main block was complete.

Back Panel Design

The back panel was easy.  I profiled a piece of 1/4″ maple to within a few thousandths of the pocket it was going into.  I pocketed recesses into the corners for the 4-40 mounting screws, drilled 1/4″ holes for the button extenders, and drilled a 5/16″ hole for the power connector.  To make room for the long power connector, I had to pocket a small recess in the back.  Some sanding and lacquer later, and it looks great.

Button Extenders

This part was fun.  To complete the all-wooden feel of the clock, I decided to use simple tactile switches to set the time, but to make button extenders out of maple so that once assembled, I’d have maple buttons.  I took a scrap maple 1×2, faced it, used a small radius tool to give the button a good feel, then profiled the button tops a few thousandths smaller than the 1/4″ holes they’d go into, and profiled the bottoms large enough to provide button stops to prevent the extenders from falling out while providing the correct amount of button travel.  The 1/4″ thick backing board they pass through was enough to provide lateral stability.  I’m particularly pleased with how they turned out.  Simple, yet effective.

Electronics Design

This clock is based on a Freescale HCS08 mcu, a Maxim DS1305 RTC with battery backup, and a pair of TI TLC5925 ICs driving an array of high brightness, wide angle white LEDs.

I’ve been a long time fan of Freescale, and I love the HCS08 series.  But, the choice of mcu variant for this project was a bit comical.  My design initially used the QG8 variant, and it worked great.  After a while, though, I realized it’s not rated for 5V and that’s the only wall wart I had laying around and was presently using.  I’m somewhat surprised that running the QG8 at 5V didn’t damage it.  The other ICs I’m using in this circuit are 3.3 and 5V tolerant, so the easy way out would have appeared to have been to just switch to a 3.3V supply.  But, I didn’t have any 3.3V regulated supplies handy at the time, so, rather than change power supplies, I changed over to the SH8 HCS08 variant.  It’s pin compatible with the QG8, so no PCB revision was needed.  Unfortunately, there is no 16-pin DIP SH8, so my board currently looks a little silly with the unused Port C pins hanging over the edge of the DIP socket.  But, I preferred the silly looking board over sourcing additional components or a board revision.

The RTC my clock uses is the DS1305.  I chose this component because it’s easy to interface, and I’ve used it before and had some S12 code that could be adapted.  Having battery backup on the RTC is a nice added feature that I get for free by choosing the DS1305 and throwing in a button cell battery holder.  I did make a minor mistake on board layout here that I’ll probably leave uncorrected.  I routed a signal line under my RTC crystal which is typically discouraged.  But, it seems to keep the time fine, so I’ll skip a PCB revision.

When choosing how to drive my 30 LEDs, I went straight to TI’s line of LED drivers – in this case the TLC5925.  TI has an LED driver for every occasion and they are very easy to drive and chain due to their use of a simple shift register for data input.  I’ve used other similar LED drivers from TI, and have found them very easy to work with, and this variant was no exception.

LED choice was also critical for this project.  I used a high-brightness, wide-angle, white LED to give the desired effect.  The combination of wide-angle light dispersion and an approximately 1 inch distance to the maple surface gives satisfactory uniform lighting across each digit segment.

PCB design was accomplished using Eagle (my all-time favorite PCB design program), and I fabricated the board using the same router I made the rest of the clock with.  For a detailed description of the process, check out the article I wrote on Mechanical PCB Etching.

The software for the clock is straight-forward.  As this is a line-powered device, I didn’t go to any great lengths to make it particularly energy efficient by using stop or wait modes.  I just enter an infinite loop which checks the two button inputs, increments the hours or minutes if requested, and updates the display when the time changes.  I also check a couple dip switches for some basic options.  I can choose between 12 or 24 hour modes, and I have a light test setting so I can verify all my LEDs are working.  None of the code is particularly clever, but it works and makes for a good basic clock.

Conclusion

The project was an enjoyable challenge.  The wood-working turned out to be trickier than I expected, and I’m probably better for it.  If you’d like to give it a try for yourself, have at it.  Here are all my design files, free for your use.  If you use them, all I ask is you don’t sue me if something bad happens, give credit where due, and if you run into me some day you buy me a beer and tell me about how it all went.  Above all, have fun.

Brad Hanken

Design Files

Main Block
SolidWorks 2010 Format
BobCad V23 Format
DXF Format
G-Code (EMC2 Post)

Back Panel
BobCad V23 Format
DXF Format
Top G-Code (EMC2 Post)
Bottom G-Code (EMC2 Post)

Button Extenders
BobCad V23 Format
DXF Format
G-Code (EMC2 Post)

PCB (Eagle 5.7.0 Format)
Schematic
Board

HCS08SH8 Software
CW10 Project Files